Big Data in E-Commerce: Investments With the Highest Return

Before you landed here, you might have already read many articles on big data that sounded more like panegyrics. Our aim is not to sing big data’s praises, but to take an unbiased look at its possible use cases in e-commerce. We will assess which of them bring the highest return and consequently should become a company’s priority.

To assess each big data application, we’ve chosen three categories – difficulty of implementation, required investments, and expected benefits – and awarded each category with points ranging from zero to 10, where 10 means the hardest case to implement, the biggest investments and the maximum benefits. The scoring results go after each use case section.

Recommendation Engine

A recommendation engine is an absolute must-have for an e-commerce company. Online store owners can hardly find a better tool for cross-selling and up-selling.

For a recommendation engine to work, it’s necessary to tune up the analytical system so that it could analyze all the actions of a particular customer: product pages they visited with the time spent there, products they liked, added into their carts and finally bought/abandoned, etc.

The system can also compare the behavior pattern of a certain visitor to those of the other visitors. The result is splendid – the analytical system works autonomously, analyzes, recommends the products that a visitor may like. More than that, the system constantly learns on the analyzed patterns and becomes even more precise over time.

Difficulty of implementation: 3

Required investments: 4

Expected benefits: 10

Personalized Shopping Experience

Creating personalized shopping experience is a key to successful e-commerce marketing. To do it, a company should be able to react to their customers’ actions properly and in real time. This becomes possible with big data tools that analyze all customer activities in an e-shop and create a picture of customer behavior patterns. Here are some good examples of how it works.

Everything in the cart is tracked (SimonVera/Shutterstock)

• A customer has put a gown, a pair of shoes, and a clutch to her shopping cart, but decided to abandon it for some reason. The analytical system knows that this customer is valuable – she shops frequently and buys a lot. Reacting immediately and offering a coupon for a 5% discount for the shoes the company may encourage the customer to finish the shopping.
• A customer bought a winter coat two weeks ago and visited some product pages with winter gloves, scarfs, and hats at that time. It’s likely that the customer will be happy to receive a personal email that advertises a new collection of winter accessories and/or announces a 10% discount for them. This may encourage him to choose your offer among multiple similar options.

Difficulty of implementation: 6

Required investments: 4

Expected benefits: 7

Voice of the Customer

Big data can help optimize the product portfolio of an e-commerce retailer. To do this, add sentiment analysis to the standard approach of analyzing products and brands by their sales value, volume, revenues, number of orders, etc.

Sentiment analysis is the evaluation of comments that the customers left about different products and brands. The analytical system automatically identifies whether each comment is positive or negative. For instance, comments that contain such words as happy, great, recommend or satisfied will be marked as positive. At the same time, a comment mentioning something bad, terrible or disgusting will be considered negative. As simple as that. By the way, the system is smart enough to mark a comment containing a phrase such as I am not happy as negative too.

If negative remarks dominate to describe a certain product, it’s a sign for an e-commerce retailer to react, for example, to get the product out of the range and send personal emails to dissatisfied customers with apologies and discount coupons for the next purchase.

Difficulty of implementation: 3

Required investments: 4

Expected benefits: 6

Dynamic Pricing

Big data can help e-commerce retailers keep in line with their pricing strategy. The concept of dynamic pricing implies setting price rules, monitoring competitors and adjusting prices in real time.

For instance, Retailer A decides that the Companies A, B, and C are their its competitors. For similar products, the retailer wants to have a price that is always 5% lower than their competitors set. Say, Company A decreased their price for 12.6 ounces of Shampoo X from $10 to $9. Our retailer’s analytical system instantly registers the change, checks it against the current price and rules and changes the price to $8.55.

By the way, an e-commerce retailer may choose to set the difference not as a percentage, but as a value (always $2 less, for example). Besides, they can create rules for the upper and the lower limits.

Difficulty of implementation: 6

Required investments: 5

Expected benefits: 6

Demand Forecasting

E-commerce retailers can significantly improve demand forecasting by creating customer profiles and looking at their customer’s behavior – when they prefer to shop, how many items they usually purchase, which products they buy, etc.

This big data needs a serious approach to collecting, analyzing and visualizing the analysis results. Besides, e-commerce retailers will analyze external big data (social media, weather forecasts, official statistics, etc.)

For instance, they can analyze weather conditions. When winter is expected to come earlier, customers may rush to buy or renew their warm clothes in October instead of November.

Let’s also imagine that a famous actress appeared in public wearing a sundress, a broad-brimmed hat and silk gloves, and this appearance triggered an active discussion on social media. The actress was unanimously recognized another icon of style. Definitely, the demand for sundresses, broad-brimmed hats, and silk gloves will increase soon. And online retailers have to be ready with these products in stock and in sufficient quantity.

An additional benefit of accurate and timely demand forecasting is that it has a positive effect on retailers’ internal processes, such as avoiding out-of-stocks, optimizing supply chain and warehouse and more.

Difficulty of implementation: 8

Required investments: 8

Expected benefits: 10

Bringing It All Together

Big data is still a relatively new concept that is not completely established yet. With that in mind, business decision makers are still concerned whether it has any practical use. That’s why we’ve focused here on describing real-world applications of big data in e-commerce.

Of course, any budget is limited and one should invest wisely to ensure high ROI. We’ve provided a scoring system that should help our readers to assess different big data use cases from the perspective of required investments vs. expected benefits. Of course, this scoring is based on the opinions of a particular big data consulting team, but we hope that it will be useful for our readers to set their priorities in their big data projects.

About the author: Alexander Bekker is the head of database and the BI department at ScienceSoft. With 18 years of experience, Alexander focuses on BI solutions (data driven applications, data warehouses and ETL implementation, data analysis, and data mining) in retail, healthcare, finance, and energy industries. He has been leading such large projects as private labels product analysis for 18,500+ manufacturers,global analytical system for luxury vehicle dealers, and more.

Related Items:

Why Big Data is the Future of Ecommerce, and Why the Future is Now

Four Common Mistakes in Machine Learning Projects

Harnessing the Power of Big Data to Reverse Retail Fortunes